JP2015119718A - Method of selecting aptamers - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for the identification of one or more aptamers to at least one target molecule.SOLUTION: The method comprising: selecting candidate aptamer sequences that bind to a target molecule, assigning to the bound sequences a measure (fitness function) of each sequence's aptameric potential, allowing evolution of some or all of the sequences to create a new mixture of candidate sequences, and repeating the method with the newly created candidate aptamer pool until the aggregate aptameric potential of the candidate pool reaches a plateau, wherein sequences present in the final pool are optimal aptamers to the target molecule.

Description

  The present invention relates to the field of aptamers. In particular, the present invention relates to aptamer libraries and methods for generating protein-specific aptamers used in proteomics such as protein biomarker identification.

  Aptamers are short polymers (usually nucleic acids (DNA, RNA, PNA)) that form well-defined steric shapes that allow the binding of target molecules in a manner that is conceptually similar to antibodies. . Aptamers are a combination of small molecules and optimal characteristics of antibodies, including high specificity and affinity, chemical stability, low immunogenicity, and the ability to target protein-protein interactions. . Aptamers have a very high affinity for the target in addition to high specificity. Typically, aptamers generated against proteins have affinities in the picomolar to low nanomolar range. In contrast to monoclonal antibodies, aptamers are chemically synthesized rather than biologically expressed, providing significant cost advantages (8, 9).

  Aptamers are typically described in U.S. Patent Application No. 07 / 536,428, U.S. Patent Application No. 5,475,096, and U.S. Patent Application No. 5,270,163. It is produced through an in vitro evolution process called “Systematic Evolution of Ligands by Exponential enrichment” (SELEX). The SELEX process includes selection from a mixture of candidate oligonucleotides using the same general selection scheme, and stepwise repetition of binding, splitting and amplification, and any desired binding affinity and binding selectivity Achieve the standard effectively. Starting from a mixture of nucleic acids, preferably comprising a segment of randomized sequence, the SELEX process involves contacting the mixture with the target under conditions favorable for binding, specifically removing unbound nucleic acid from nucleic acid bound to the target molecule. Splitting, dissociating the nucleic acid-target complex, amplifying the nucleic acid dissociated from the nucleic acid target complex to yield a ligand-enriched mixture of nucleic acids, and binding, splitting, dissociation and amplification steps Is repeated for the desired number of cycles to obtain only the sequence with the highest binding affinity for the target molecule. Candidate oligonucleotides can contain fixed motifs or known motifs within the sequence. If purely random sequences are used, the discovery of selective aptamers within the candidate population will be entirely by chance. In fact, the more random the oligonucleotide sequence, the more likely the choice of that sequence for the target under study will depend on chance (11, 12, 13).

In its most basic form, the SELEX process can be defined by the following sequence of steps.
1) Prepare a candidate mixture of nucleic acids of different sequences. A candidate mixture generally includes a region of fixed sequence (ie, each member of the candidate mixture contains the same sequence at the same position) and a region of randomized sequence. The fixed sequence region may either (a) assist the amplification process described below, (b) mimic a sequence known to bind to the target, or (c) a given number of nucleic acids in the candidate mixture Selected to enhance the concentration of the structural sequence. Randomized sequences can be completely randomized (ie, the probability of finding a base at any position is a quarter), or only partially randomized (eg, the probability of finding a base at any place is Can be selected at any level between 0 and 100 percent).
2) The candidate mixture is contacted with the selected target under conditions favorable for binding between the target and a member of the candidate mixture. Under these circumstances, the interaction between the target and the nucleic acid of the candidate mixture can be judged as the formation of a nucleic acid target pair between the target and the nucleic acid having the strongest affinity for the target.
3) The nucleic acids with the highest affinity for the target are split from those nucleic acids with the lower affinity for the target. Only a sequence corresponding to a very small number of highest affinity nucleic acids (and possibly only one molecule of nucleic acid) is present in the candidate mixture, so a significant amount of nucleic acids in the candidate mixture (approximately 5-50%) However, it is generally desirable to set the split criteria so that it is maintained during splitting.
4) Next, amplify the nucleic acids selected during the split as having a relatively higher affinity for the target to produce a new candidate mixture enriched for nucleic acids having a relatively higher affinity for the target. Generate.
5) By repeating the resolution and amplifying the above steps, the newly formed candidate mixture will contain less and less weak binding sequences, and the average degree of nucleic acid affinity to the target will generally increase. In the extreme case, the SELEX process will result in a candidate mixture containing one or a few unique nucleic acids representing those nucleic acids from the original candidate mixture having the highest affinity for the target molecule (11, 13).

  One of the main problems when generating an aptamer library is the overall size of the possible search space. For example, there can be hundreds or thousands of possible oligonucleotide sequences that can be selective and specific for a single protein, but where to start to find the most correct sequence?

The term “search space” encompasses all possible or permissible variations of polymer units (such as nucleotides) that can occur in an aptamer molecule of a given length. Since only a limited number of sequences can be queried at a time, the candidate library can only sample a fraction of the possible search space for possible nucleotide sequences of a given length. For example, 40 mer is ^{10 24} possible including four basic units (nucleotides for DNA / RNA). Thus, the aptamer search space (for a 40mer nucleic acid) relative to a typical protein space from a biological sample (assuming it contains tens of thousands of different proteins) is -10 ²⁰ : 1 (± 10 times wider) ) In the ratio of candidate aptamer polymer to protein. At 60 mer, it is 10 ³² : 1. Even if only one of the 10 ¹⁰ sequences is strongly “aptameric”, it still leaves a 10 ²² : 1 aptamer-pair protein that is queried to find the sequence of one strong aptamer.

  However, there can be a large number of aptamers of various qualities for any protein. SELEX thus relies on the appropriate sequence being present in the initial candidate mixture under test by chance. The method then converges exponentially to the most correct solution available in the selected mixture. Some prior knowledge of possible suitable sequences (eg, sequences that give a particular secondary or tertiary structure to a sequence) ensures that the process is not completely determined by chance Can be used during candidate library design to ensure. Even so, the solution provided by SELEX is very unlikely to provide the most correct solution for the problem (ie, the most accurate aptamer for a particular protein).

Thus, the SELEX process favors the use of a single expressed and isolated protein target applicable for binding selection, through a small set of aptamers or rounds of amplification towards convergence to a single aptamer. Progress randomly. The process finds the “most correct” aptamer for a protein isolated from a given set of sequences by exponential and iterative selection. In recently published schemes, often only 1 × 10 ^{−20th of} the possible sequence search space is utilized (8), but certain basic sequence motifs are much more aptamers than others. It is understood that Nevertheless, current aptamer searches are unlikely to find the best fit aptamer from the pool overall. Also, the search results are very dependent on the initial pool diversity. In most current SELEX schemes, aptamers must be in the initial search pool of candidate aptamers. Each protein is also very likely to have a broad spectrum of possible aptamers within the complete aptamer space, considering the small searches made to date. Since the SELEX scheme is exponential and generally a putative aptamer must be present in the initial library, selecting aptamers for multiple protein targets simultaneously, or in the background of others Selecting aptamers for a single protein target masked with SELEX would be difficult with SELEX. SELEX is also unable to fully screen 10 ¹⁵ candidate aptamer sequences recently claimed to be typical in the literature. This is because the quality of library synthesis on a commercially available DNA / RNA synthesizer cannot be controlled to the level necessary to demonstrate library diversity. Also, some of the aptamers in the library are very likely to anneal (stick) to others, some of which are stoichiometric in the population due to bias in the synthesis process. Will be expressed less. Many aptamers will fold into a tertiary structure distribution (some are active and others are inactive). Many will simply have no opportunity to bind to the target protein due to their relative dilution. Thus, 10 ¹⁵ candidate aptamers are unlikely to be actually screened against the protein, so the search is more limited than possible and programming that relies on a means to generate random sequences. Several variations on the SELEX scheme have been described that attempt to circumvent these limitations by introducing random variations into evolving aptamers (7).

  In another aspect, SELEX is essentially a physical in vitro embodiment of a computer heuristic. A heuristic is an approximation method for assisting in solving a problem in time with a “reasonably good” correct answer. In this case, the problem is to find the most correct solution from a finite search space that is too large to search as a whole. Heuristics are computer-based methods that use trial and error methods that approximate solutions to problems that are difficult to solve on a computer. In other words, a heuristic method or procedure starts with only an approximation method that solves the problem within the context of some goal, then improves its own performance and therefore moves towards a better solution Use feedback from the effects of the solution. SELEX performs a basic search for randomly generated candidate aptamer sequences against the target program, assesses the success of each candidate sequence result by its survival rate in the analysis, and then further rounds of reselection. Amplify the residue before. Eventually, the strongest candidate sequences from the initial library (ie those that bind to the target protein with the highest affinity) become progressively more concentrated in subsequent rounds of the procedure. The SELEX procedure is therefore a laboratory run of a computerized procedure to find the polymer (or nucleic acid) with the highest binding affinity for the protein under study. If the physicochemical properties of aptamers and proteins and the causes of their behavior are all known, it would be possible to model the SELEX procedure in a computer in a completely deterministic manner. This process can be approximated with less complete information (17). Further examples of the use of biomolecules that solve computer problems more clearly have been published (1, 2, 3, 4, 5).

US patent application Ser. No. 07 / 536,428 US Patent Application No. 5,475,096 US Patent Application No. 5,270,163

  In order to solve this problem and improve the search capabilities and solutions presented by SELEX, the inventor explored the aptamer search space and the numerous proteins present in the proteome and a large number of possible for each protein We recognized the need for an intelligent method to find the optimal solution to deal with the scale of the problem presented by the candidate aptamer.

  The evolutionary growth of a group of candidate solutions as a means to find the correct answer to a very complex problem with a large number of possible solutions has been well described by computers (14, 15, 16), so-called genetic algorithms Best embodied in (a specific class of evolutionary search heuristics). Genetic algorithms are search algorithms based on natural selection and natural genetics mechanisms (growth, mutation, recombination, natural selection and survival of the fittest). They combine structured but still randomized information exchange with survival of the fittest in a coded representation of the candidate solution (eg string structure). This makes it possible to form a search algorithm with some of the innovative intuitions of human search. Candidate solutions to the optimization problem under study play the role of artificial creations (individuals) in the population. Next, population evolution occurs after repeated application of the above operators (mutation, crossover, propagation and selection). In all generations, a new set of artificial creations (string coding or binary coding) is generated using pieces of the old or previous population in pieces. Sometimes a new part is tried for correct measurements. The genetic algorithm efficiently utilizes past information to infer about new search points with the expected improved performance (14).

  In these schemes, the search problem is encoded by each possible solution s expressed as a chain of numbers or letters, a sequence or a string. This encoding is for the convenience of heuristic execution only. These chains, sequences or strings of numbers or letters are often described by analogy as “genome”. Each artificial individual solution thus has its own genome. Many initial populations of candidate solutions (genome) are generated, but the population represents only a very small portion of the total number of possible candidate solutions that can be fully encoded. These heuristics are typically developed when an approximate correct answer is too large to be searched from a search space that cannot be handled by an exhaustive search. The “fitness function” or “objective function” f is applied to each possible solution s or member of the candidate population. This function is an evaluation of the proximity of the individual solutions to the optimal solution features. A high f (s) suggests that s is the correct solution.

  Usually, the initial group of randomly generated candidate solutions s includes the first generation. The fitness function f is applied to the candidate solution and any subsequent descendants. In selection, the parent for the next generation is selected with a bias towards higher fitness. Only “adaptive” individuals, as measured by application of the fitness function f and a cutoff (eg, upper quartile), are selected for the second round of mutations (random changes to their genome). And crosses a portion of their “genome” with other high-scoring individuals (recombination or propagation). The method used to do this is to change the algorithm of those “genome” units from one possibility to another, and the genome prefix at random points between any two adaptive genomes. And the production of postfix combinations. Thus, a second generation of children is generated in addition to the original parent solution. Expressed in more detail, parents grow by copying by recombination and / or mutation. Proliferation is a process in which individual candidates or strings are copied according to a fitness function f. This is because a candidate with a higher fitness value is more likely to give one or more offspring to the next generation, and some of the internally coded features are transmitted to the offspring It means to become. This operator (f) is an artificial version of natural selection (Darwinian survival of candidates among candidates). Once a candidate is selected for growth, an exact replica of the candidate is created. This candidate is then entered into a mating pool (temporary new population) for further general operator action.

  Recombination operates on two selected parents (candidates) resulting in one or two children (new candidates). After growth, a simple crossover can proceed in two steps. First, new candidate or child members in the mating pool are arbitrarily mated. Second, each pair of new candidates crosses to generate two more new candidates.

  Mutations work on one candidate, resulting in a new candidate. Mutations are necessary even if effectively searching for and recombining surviving growths and crossings can accidentally become excessive and cause some useful potential genetic material to be lost. . In artificial systems, mutation operators protect against such irreversible reductions. Mutations are therefore insurance against early loss of important things.

  In summary, these operators generate offspring (a set of new candidates), resulting in a new set of diverse but similar individual candidate solutions (or genomes) of several arbitrary population sizes. This new set is assessed for its fitness and the most correct solution survives to grow and mutate. These new candidates will compete with their old candidates in the next generation (surviving the fittest). This is done until the total fitness of the population reaches a stable plateau. Often this means that there are several individuals that are reasonable solutions to the search, and often they have some features in common. Thus, a group of candidate solutions (they are coded as they are) can converge towards a set of solutions that meet a given “fitness” criterion. The resulting population of individuals may have little compositional commonality with the initial population. The number of such iterations and the number of considerations necessary to achieve a stable and highly adaptive candidate solution population is relatively small and much less than an exhaustive search.

  Thus, genetic algorithms allow iterative searches for solutions or sets of solutions for problems in space that are too large to search all at once. The problem to be solved is first coded in a set of candidate solutions, and a fitness function is applied to each possible solution. The most correct solution can then be identified. The heuristic then generates a diverse but similar group of “children” that possess their parental features. The fitness question is then reapplied and the process is repeated. Thus, starting with a random approximation of the solution to the search, the algorithm allows the movement of the population to the optimal set side of the solution, keeping in mind that the most correct solution is dependent on the robustness of the fitness function. To do. While there are multiple equally correct solutions for a given problem, the strength of genetic algorithms tends to find some or all of these equally correct solutions under the right conditions. That is.

  With this in mind, the present invention encompasses improvements in SELEX. In other words, the present invention relates to the application of computerized heuristics to improve SELEX. For this purpose, the present invention is a physical embodiment of a genetic algorithm. Indeed, in one aspect, the invention is the use of a genetic algorithm paradigm that directs polymer sequence evolution in the design of candidate aptamer sequences.

  In another aspect, the invention is a method for the identification of one or more aptamers to at least one target molecule, including SELEX, which is directed to the best-fit sequence for each target molecule. The method is characterized in that it further includes the use of fitness of candidate aptamer sequences to direct sequence evolution.

Specifically, the present invention is a method for the identification of one or more aptamers against at least one target molecule comprising:
a) selecting candidate aptamer sequences that bind to the target molecule;
b) assigning to the combined sequences a measure of the aptamer potential (fitness function) of each sequence;
c) enabling evolution through random or directed changes to some or all of the sequences to generate a mixture of new candidate sequences;
d) repeating steps a) -c) with the newly generated candidate aptamer pool until the total aptamer potential of the candidate pool reaches a plateau,
This is a method in which the sequence present in the final pool is the optimal aptamer for the target molecule.

More specifically,
a) contacting a pool of candidate polymer sequences with at least one target molecule;
b) splitting the unbound sequence from the sequence specifically bound to the target molecule;
c) dissociating the sequence target complex to obtain a mixture enriched in ligands of the sequence;
d) assigning each sequence obtained in step c) a measure of the aptamer potential of the sequence (fitness function);
e) using the measurements of step d) to determine the aptamer potential of the ligand enriched mixture;
f) using the information obtained in step e) to allow the evolution of some or all of the sequences obtained in step c) to generate a new sequence mixture;
g) repeating steps a) -f) with the newly generated candidate aptamer pool until the total aptamer potential of the candidate pool reaches a plateau,
A method in which the sequence present in the final pool is the optimal aptamer for at least one target molecule.

  Thus, the method of the invention allows the identification of aptamers that are selective for the target (substantially the most correct aptamer). The candidate polynucleotide sequence of the aptamer itself represents a possible solution within the search space (the search space is all possible aptamers (and implicitly their sequences)). The use of sequence evolution allows a population of possible solutions to move towards an optimal solution. Thus, the sequences in the final pool did not have to be present in the original search pool, and in fact were probably hardly present in the original search pool. Furthermore, because SELEX queries a fixed population of possible solutions, the opportunity to succeed in obtaining the most correct solution (aptamer) by allowing the evolution of candidate sequences in the pool is more likely than SELEX. Is expensive.

  The term evolution encompasses sequence propagation, recombination, crossover and mutation during the selection process iterations. Evolution can be random, designed, or a combination thereof.

  It should be noted that since the only thing that distinguishes one polynucleotide aptamer from another is its sequence, the properties of that aptamer must therefore be encoded in that sequence.

  The assignment of aptamer possibilities is done using one or more measured or calculated properties specific for each sequence. For example, aptamer potential can be based on the abundance of sequences in a mixture enriched for ligands (relative abundance is compared to previous repeats or controls). Quantification can be combined with other measures of aptamer potential. Such measurements are typically derived from the sequence itself and properties given by the sequence such as secondary or tertiary structure predictions, hydrophobicity, similarity to known aptamers, and the like. Hybrid measurements that combine or aggregate these measurements mathematically may also be appropriate. The possibility of a particular general sequence motif can also be used to estimate the aptamer potential of a sequence. The method of determining the relative or absolute aptamer potential of a candidate sequence from its measured statistical or sequence features (or others) is called the “fitness function”. This follows standard terminology for those skilled in the evolutionary search heuristic (14). Other terms also exist for the same concept, such as “objective function”. Expressed non-technically, this is an objective measure of the potential of a candidate aptamer (usually a score).

  Features of the present invention allow individuals who are highly aptamers to undergo small random compositional changes such as a single change in sequence and cross-recombination of sequences, thereby creating stronger candidate aptamers. Create a new pool of arrays. A significant advantage of allowing sequence evolution is that it allows the generation of sequences and motifs that cannot exist in the initial candidate pool, and thus the possibility of identifying the optimal aptamer for the target protein Promote. Another significant advantage of this method is that only “candidate” candidate sequences are allowed to recombine with each other, thus generating a putatively more adapted child population of aptamer sequences. Is to do.

  The basic difference between the scheme of the present invention and SELEX is that SELEX does not readily allow for the introduction of mutations or changes in the sequence under study. In fact, the only way that mutations can occur is by chance during the amplification process when the combined sequences are amplified by PCR. However, such mutations are so rare that they may not affect the process (ie the overall composition of the population of sequences under study). SELEX also does not activate or allow recombination or crossover of sequences between aptamer species (otherwise referred to as “recombination or propagation”) or rational intervention to alter aptamer sequences. . Furthermore, the SELEX scheme does not allow sequences to be assessed for their “fitness” as aptamers, for example, and manipulated rationally between rounds of selection.

  An example of a reasonable intervention may be to ensure that all aptamer sequences at any repeat contain a particular sequence motif. This may not occur after mutation and recombination. Thus, reasonable intervention acts as a filter or constraint on randomness. Importantly, prior knowledge (eg, known protein binding motifs) can be maintained, eg, incorporated into a population of sequences under study, thus improving the efficiency of the selection process. it can. This does not fully determine the outcome, but allows humans to direct and constrain evolution.

  In a further example, monitoring the sequence pool diversity at each iteration can help the population find multiple solutions (aptamers) for multiple proteins in parallel. This can be accomplished by identifying, monitoring and directing the evolution of multiple subpopulations of distinct sequences within the entire population. Unlike the proposed invention, SELEX does not allow the quantification of bound aptamers at each iteration of the process, and for each iteration of the aptamer pool in a manner that positively affects the overall results. This is not possible with SELEX since this information is not used to make inferences and decisions about each sequence in between.

  The so-called “next generation” polynucleotide sequencing technology involves separating individual molecules of DNA or RNA or other polynucleotide sequences onto a surface such as a bead or chip, thereby producing a single molecule array of sequences. I need. The array has a surface density that allows each molecule to be resolved individually, for example by optical microscopy. Sequencing of the polynucleotide molecules on the array allows a “digital” (ie, absolute) count of sequences, and thus direct quantification of the sequences present in the array. In some techniques, sequences can be arrayed once and clonally amplified to enhance and / or define the signal from each sequence. Nevertheless, quantification is obtained by counting the occurrence of sequences on the array, not the signal produced from each amplicon. Examples of suitable sequencing and quantification techniques can be found in publications such as WO 00/006770 and Branton et al. (21).

  In the proposed invention, manipulation of candidate aptamer sequences is accomplished through the use of massively parallel DNA sequencing, such as techniques embodied in “second generation”, “next generation” or “third generation” DNA sequencers. Is done. The use of massively parallel sequencing allows candidate aptamers enriched for ligands to be quantified and sequenced in parallel, thereby adapting fitness measures derived from their abundance, and simultaneously to their sequences Provides information on other fitness measures from which it comes. In view of the performance of massively parallel sequencers, such a scheme will significantly reduce the time and cost of experiments to find aptamers under the proposed scheme.

  Thus, another example of rational intervention (or human orientation) is to examine and count the aptamer sequences under study, pointing out the disadvantages of the sequencer itself that impairs the display of specific sequences. The This can be compensated by biasing the composition of the subsequent population.

  Thus, in a preferred embodiment, the aptamer potential is measured by quantification of each sequence obtained in step c). Ideally, quantification is performed using a single molecule array of sequences, or similar devices that can sequence and count individual molecules in a massively parallel fashion. In particular, sequencing or partial sequencing sufficient to identify each sequence on the array is performed in combination with a count of each sequence to achieve quantification. Alternatively, each sequence can be amplified after being arrayed on the surface and the sequence quantified on an array of clones. In this way, the primary fitness measure can be a count on the array that represents the frequency of a given candidate aptamer sequence in the population under study. Once primary fitness is obtained, bioinformatics data (such as similar motifs or secondary structures) derived from the nucleotide sequence or molecular composition of each candidate aptamer and obtained during quantification is Can be used in addition to drilling down into enriched sequences to obtain additional fitness criteria. The calculated but composition-dependent properties continue in the evolutionary generation of candidate aptamer libraries for the next round, either by incorporating them, biasing them, or eliminating them. Can be used.

  The present invention takes advantage of the capabilities of techniques currently available and in a massively parallel fashion to separate and quantify individual sequences to single molecule sensitivity. It also allows the study of complex and diverse populations of sequences. Candidate aptamer sequences that are found to be more abundant in the bound fraction when screened against a single protein are the fact that the method searches for candidate aptamer sequences that adhere (bind) to the protein under study Thus, it has a higher fitness than a sequence in which only one copy is combined. This is the primary characteristic that defines aptamers. When screened against complex mixtures of proteins, other statistical properties derived from the quantification procedure are required to identify aptamers. The advantage of massively parallel sequencing is that it allows each and every sequence to be counted and thus quantified. This is in contrast to, for example, antibody analysis where the resolution cannot be as high.

  Thus, aptamer counts, and other measured, calculated, or historical statistics generated from their sequences can be used and combined into a “fitness function”. This allows the physical implementation of evolutionary search heuristics that are refined and more successful at the molecular level than those embodied in SELEX. Like such sophisticated heuristics in the computational domain, it is important research to derive and apply such “fitness function” itself rather than details of a given “fitness” measurement itself. The ability to improve subsequent iterations of the population of molecules below. It should be noted that the present invention allows many such investigations, thereby allowing learning of the optimal general “fitness” characteristics of aptamers, or context specific “fitness” for a particular class of proteins, for example. ”Will be derived.

  As information about candidate sequences and motifs is constructed, knowledge of particular features can be incorporated into later candidate pools or new pools for identification of aptamers to related or similar proteins. Similarly, sequences and motifs found to be ineffective can be actively excluded from the candidate pool.

  Also, if aptamers with selectivity for different proteins are required, known aptamer sequences for known proteins can be actively excluded from the candidate pool. This helps to ensure that the aptamers found are likely to be specific for this protein (a major requirement for useful aptamers). Alternatively, if sequences and / or motifs are selected that provide selectivity and specificity for a particular protein family, these sequences and / or motifs are then used to identify aptamers that are selective for these protein subtypes. Can be incorporated into the candidate aptamer pool used in

  All of these advantages include knowledge of aptamer sequences, their relative “fitness”, as well as other computable features derived from their sequence or their quantitative behavior under selective schemes. Is promoted by having

  The algorithm encoding scheme used by the genetic algorithm reflects the actual compositional nature of the present invention. These algorithms encode the search problem into a “genome” with an analogy of the nature or number (or bits) of iterative changes such as cross-recombination and selection / amplification, even in computer memory To do. In accordance with the present invention, these processes are performed using the actual molecule under study. The molecules (polynucleotide sequences in this case) themselves encode possible solutions to the search and their properties such as their stickiness to proteins (as represented by quantification) (also in their sequences) Coded), which drives the method (15). Thus, the present invention is a physical embodiment of a genetic algorithm. This is made possible by the use of massively parallel DNA / RNA sequencing and the fact that aptamers are individual polymer molecules whose aptamer properties are encoded in the sequence.

  More specifically, the use of massively parallel sequencing allows the execution of optimized semi-rational / semi-random evolutionary search strategies for the identification of specific aptamers against protein targets or multiple protein targets To. For example, known sequence motifs previously shown to be effective against proteins similar to those under study are preferred both as part of initial library generation and as part of fitness selection. Can be selected. In this way, essentially and conveniently towards a high-quality solution or a “fit” solution that includes these motifs, which are also a precondition for any solution, Bias the process. However, sufficient randomness in the library generation and subsequent mutational modification of the motif can be used to ensure finding new but similar sequences that bind the protein under study. The fitness function can be designed to select sequences that are similar but sufficiently different from those already found for the previous protein, and both success and specificity in selecting the appropriate aptamer. Guarantee. This also facilitates the accumulation of a catalog of aptamers that are sensitive and specific to many proteins over time. Such “non-overlapping” subpopulations of distinct aptamers that operate on one protein and not the other can then be used simultaneously to probe and measure multiple proteins in parallel. .

  Since massively parallel sequencing allows multiplexing, aptamer libraries for more than one protein can be generated and screened simultaneously. Similarly, the present invention allows many aptamers of high quality but different (at the sequence level) to be selected in parallel against single or multiple protein targets.

  Furthermore, since the massively parallel sequencing allows quantification of each sequence species, and each species can represent a single protein, the proposed method uses genomics to reduce the power and dynamic range of proteomics. Expand.

  In a specific embodiment of the invention, a diverse library of designed, random, or semi-designed sequences can be used for folded proteins derived from biological material (eg, serum, plasma). Screened and selected against in vitro set. The library of sequences may be generated by any method.

  Sequences that do not bind protein are removed or eluted. Various known options are available to accomplish this. For example, a single protein or complex protein sample can be immobilized on a solid support. Stringent washing can be performed to remove unbound and weakly bound sequences. Another option is the use of reversible cross-linking (photoaptamers) of sequences on protein targets.

  The remaining binding sequences are removed from their protein host and passed through a massively parallel sequencer when they are sequenced and counted.

  As a specific example of massively parallel sequencing, binding sequences are randomly arrayed on a surface such as a bead or chip. The sequence is optionally cyclically amplified, resulting in a group of clonal single-stranded molecules at individual x and y coordinates on the surface or on individual beads. A suitable DNA sequencer then performs stepwise chemistry consisting of reagents that allow the determination of one base on each complementary sequence per cycle to monitor base incorporation in real time. The illumination and imaging system allows this process to be filmed so that the original candidate sequence can be obtained. Regardless of the details of the sequencing technique, a complementary sequence that reflects each combined candidate sequence is constructed. Typically, such techniques can be sequenced at a maximum length of 75 base pairs beyond 40 million to 300 million DNA fragments, and these numbers are rapidly increasing as the technique is improved. . This process currently takes less than 1-3 days from sample preparation to sequencing output, and these time scales are shortening.

  It should be noted that other massively parallel methods that fall into the category of “next generation polynucleotide sequencing” can be used and the invention is not limited to the specific examples described. “Next generation polynucleotide sequencing” is a term made to describe a DNA / RNA sequencing platform that generally emerged in 2004. Since 2008, another generation of sequencing platforms with improved features is now represented as “third generation”. Their common feature is the use of different sequencing chemistry than the old technology based on "Sanger" sequencing. New platforms use new chemistry and are typically of very high throughput and much lower cost. This was achieved through the ability to parallelize sequencing reactions to a very high degree. Unlike the Sanger method, where new platforms function by truncating (degrading) bases, they typically progress through synthesis or chain extension (construction), but are not limited to this. In addition, the new platform operates with a small number of single or clonal molecules, unlike the Sanger method, which has a DNA sequencer with a very high molecular to molecular ratio.

  Although DNA sequencing is mentioned, DNA sequencing techniques can also be used for RNA or PNA (peptide nucleic acid) sequences. Thus, references to next generation sequencers or sequencing encompass all other chemical variants of nucleic acid based polymers suitable for use in the methods of the invention and their analogs as well as DNA, RNA and PNA. To do. All new sequencing technologies can sequence individual molecules or their clonal copies in a massively parallel and highly efficient manner, whether they are “next generation” or “third generation”. it can.

  The use of “sequence counts” (ie, quantifying the absolute or relative abundance of sequences in complex biological samples (such as miRNA)) is already well established, and such next generation platforms or A massively parallel platform is described (18). On an effective massively parallel sequencing platform, derived sequences complementary to the original candidate sequence are highly accurate, eliminating homopolymer and palindromic sequences or other motifs that are strong structural elements There should be little, if any, significant systematic sequence context bias that cannot. On some platforms, this has already been established by resequencing a large complex genome, which contains such a motif.

  The methods of the invention can be performed on a single isolated protein. Alternatively, the method can be performed on a single protein known to exist in a mixture of proteins.

  However, one of the advantages of genetic algorithms is that subpopulations can evolve from a wider population. This is achieved by sub-selecting against different criteria. Thus, in a further alternative, the method also allows interrogation of many proteins in the mixture at the same time. This technique allows for multiplexing, which is made possible by the use of next generation DNA sequencing techniques.

  When searching for candidate aptamers for multiple proteins, the appearance of a distinct population of candidate sequences can be monitored. The sequence population can then be classified and developed in parallel under the general scheme in the present invention. For example, distinct aptamers for different regions of a single protein can be identified. Note that the number of populations queried at any time is limited by the dynamic range of the target molecule in addition to the sequencing array.

  Although the methods of the invention can be used to query a single target or protein (eg, a protein excised from a gel), it is particularly suitable for analysis of a mixture of targets, including complex protein mixtures. The term “protein mixture” or “protein mixture” generally refers to a mixture of two or more different proteins (eg, a composition comprising two or more different proteins or isoforms thereof).

  In preferred embodiments, the mixture of proteins analyzed herein is greater than about 10, preferably greater than about 50, even more preferably greater than about 100, even more preferably greater than about 500 different proteins, such as More than about 1000 or more than about 5000 different proteins can be included. Exemplary complex protein mixtures include, but are not limited to, all or a fraction of proteins present in a biological sample or portion thereof.

  The term “biological sample” or “sample”, as generally used herein, refers to material that is obtained from a biological source and is in an unpurified or purified form. By way of example and not limitation, the sample can be a virus (eg, a prokaryotic or eukaryotic host virus); a prokaryotic cell (eg, a bacterium or archaea (eg, a free-living prokaryotic or planktonic prokaryote, or Colonies or biofilms containing prokaryotes)); eukaryotic cells or organelles thereof, including eukaryotic cells obtained in vivo or in situ or cultured in vitro; eukaryotic tissues or organisms (eg, Eukaryotes can be obtained from eukaryotic tissues or cell-containing or cell-free samples); eukaryotes are protists (eg protozoa or algae), fungi (eg yeast or mold), plants and Animals (eg, mammals, humans or non-human mammals) can be included. A biological sample can thus include, for example, cells, tissues, organisms, or extracts thereof. Biological samples include collection or collection of urine, saliva, sputum, semen, milk, mucus, sweat, feces, etc., collection of blood, cerebrospinal fluid, interstitial fluid, intraocular fluid (glass body fluid) or joint fluid, etc. It can preferably be removed from its biological source (eg from an animal such as a mammal, human or non-human mammal) by any suitable method, including but not limited to, or by tissue biopsy, excision and the like. Biological samples can be further subdivided to isolate or concentrate portions thereof used to obtain proteins for analysis in the present invention. By way of example and not limitation, a variety of tissue types can be separated from each other; specific cell types or cell phenotypes are isolated from samples using, for example, FACS sorting, antibody panning and laser capture dissection Cells can be separated from interstitial fluid, for example, blood cells can be separated from plasma or serum; or the like. Samples can be applied directly to the methods of the invention or can be processed, extracted or purified to varying degrees prior to use.

  The sample can be from a healthy subject or a subject suffering from a disease state, disorder, disease or infection. For example, the subject may be a healthy animal (eg, a human or non-human mammal), or cancer, inflammatory disease, autoimmune disease, metabolic disease, central nervous system disease, eye disease, heart disease, lung disease, liver disease, It can be, but is not limited to, an animal (eg, a human or non-human mammal) with gastrointestinal disease, neurodegenerative disease, genetic disease, infection or viral infection, or other disease (s).

  Preferably, in order to increase the sensitivity and performance of proteome analysis, a protein mixture derived from a biological sample can be processed to deplete a high amount of protein therefrom. By way of example, mammalian samples such as human serum samples or plasma samples contain abundant proteins, especially albumin, IgG, antitrypsin, IgA, transferrin, haptoglobin and fibrinogen, which are preferably so depleted from the sample. It may be. Methods and systems for the removal of abundant proteins are known, such as immunoaffinity depletion, for example, Multiple Affinity Removal System (MARS-7, MARS-14) Are often commercially available from Agilent Technologies (Santa Clara, Calif.).

  Although the present invention has particular application to the identification of specific aptamers for proteins, the present invention also applies to interrogation of other molecules such as metabolites and possible small molecules and biotherapeutic agents. It should be recognized that it has.

  The method described above focuses on the quantification of aptamer sequences via sequencing. For the purpose of generating a library of specific aptamers for a single protein, it is not necessary to specifically examine the aptamer sequence until the library is complete. Thus, the method can further comprise obtaining a sequence of only the final pool of aptamer sequences.

A significant advantage of this method is that the amount of material required is small (picomoles, even fentomoles). If a 40-mer DNA aptamer sequence is prepared using 4 bases (A, C, G and T), then a possible 1.2 × 10 ²⁴ combination is available. Thus, one mole (6.2 × 10 ²³ ) may contain only 5% of all possible sequences in the best case, and such a preparation would exceed 1 kg in weight. One picomole contains approximately 1 × 10 ¹¹ molecules, so a 1 ml picomolar unique aptamer will contain 1 × 10 ⁸ molecules. If 17 base pairs in a 40mer are fixed as a “motif”, leaving 23 variable base pairs and / or random base pairs, 1 picomole contains up to about 70 copies of each possible sequence, so A 1-10 ml picomolar solution should contain between 1 and approximately 10 copies of each sequence. Thus, a quantifier (ie, a massively parallel sequencer) can and should exceed both the aptamer affinity for the protein and the native concentration of most proteins in the unmodified biological sample in sensitivity. It is.

Next generation sequencing devices such as those sold by Illumina® typically image 330 × 8 on a surface that can cleanly extract as many as 25000 sequences per imaging area. Has an area. Therefore, it should currently be possible to query about 1 × 10 ⁸ 40mer arrays per run and per chip. Therefore, this technique should allow 1: 1 observation of picomolar aptamers in solution. Thus, the technology has a dynamic range that spans the dynamic range of proteins in the native sample. The performance of such sequencing techniques has improved over time.

  Once a specific aptamer library or population for a particular protein has been developed, the sequence of the aptamer can be used to rationalize the structure, function and binding characteristics of the aptamer population for a specific protein. Knowledge can be built, or vice versa. In turn, this can be used to further investigate the protein and improve other parameters and / or features of the aptamer library. These refinements can be applied either during initial library generation and / or during a round of iterative selection under the proposed scheme. In fact, over time, sequence / structural relationships between identified aptamers and proteins are built and used as input to the design of the initial aptamer library, with different motifs and other parts of the aptamer space A part can be investigated.

  Such information can also be used to evaluate proteins having a structure similar to the initial protein of interest. In this way, aptamers for a particular class of proteins are found and similar proteins are investigated and queried by using a starting aptamer library that is reasonably selected based on prior knowledge of proteins in the same class. be able to. Can the knowledge of library sequences for one protein, a particular domain, also be used to assist in the design of aptamer pools for related proteins and other members of the hypothetical protein family, Alternatively, the ability of similar aptamer sequences to bind to the surface of similar proteins can be demonstrated and measured.

  Discarded aptamer sequences can also be mined for information. For example, the statistical and computational properties (eg, structure) of these sequences can be used to characterize the general properties of weak aptamers. When designing an initial aptamer library, such information provides useful knowledge and allows a shortcut to limiting the randomness of the initial library as a sequence level.

  In an alternative embodiment, the intermediate step is to examine aptamer tags (sequences) bioinformatically and rationally improve them, and thus a population directed towards diverse but highly specific tags Can guarantee the evolutionary progression. Bioinformatics data includes sequences that yield specific secondary and tertiary structures, and sequence complementarity between candidate sequences and proteins.

  As outlined previously, features can be calculated, aggregated, related, and stored for a given aptamer sequence and other biological data such as protein families, folding domains, etc. . These bioinformatics features can be used to improve initial library generation. They can also be used as part of a “fitness function” between selection rounds of candidate aptamer sequences to shape and improve the characteristics of the population at each stage.

  Proteins in the blood are specific targets for identification of disease state markers and drug treatment. It is widely assumed that the amount and / or conformation of proteins in the blood should be statistically related to such conditions in a manner that exceeds endogenous natural fluctuations. Blood and other bodily fluids immerse affected tissues, transport vital proteins, and can be obtained for testing using relatively inexpensive and simple procedures during medical consultations So blood and other body fluids are specific targets.

  However, proteins in the blood have a very wide range of concentrations, with a small number of proteins accounting for more than 99.9% of all proteins and the rest accounting for a picogram to milligram distribution per milliliter (19) . Thus, a small population of high abundance proteins hides the necessary but rare proteins for life that are also present in the protein mixture.

  The method of the present invention allows the identification of aptamers to proteins present in biological samples in low or very low abundance. Thus, in a further embodiment, the method further comprises removing a very high amount of candidate sequences found in the initial (first and / or second) round and / or other rounds of iteration. Given that there is a finite number of “slots” in the sequencing array, a highly abundant candidate sequence occupies most if not all slots, thereby hiding and potentially occupying any less abundant candidate sequence It will provide a distorted overview of the dynamic range of the protein. By removing very high abundance candidate sequences from the sequence pool, either in absolute or relative sense, then specific candidate sequences for highly abundant proteins are no longer This particular population of proteins within a complex mixture is effectively ignored, even though it cannot be interrogated and the protein is still present in the mixture.

  Removal of a sequence that accommodates one highly abundant protein will indicate another highly abundant candidate sequence or a candidate sequence for a smaller amount of target.

  This is another example of a rational intervention that biases in finding an aptamer sequence directed to a smaller amount of protein. Other equilibrium choices (such as low variance for putative low abundance aptamers between duplicates of the same repeat and / or preference for certain known sequence motifs) may also be required in the fitness function Become. This will help ensure that low abundance but “adapted” sequences are selected from background noise of non-specific or random sequences.

  The advantage of this additional step is that protein mixture manipulations that are very highly abundant and remove common proteins are avoided. In this way, the protein mixture is more faithfully based on natural samples, and by reasoning, any information about the protein derived from the mixture is more likely to more accurately reflect the natural status of each protein. This use of the native state of the protein also has the added advantage that it allows the selection of specific aptamers directed to proteins with chemical modifications or other post-translational modifications that are common in nature.

  Highly abundant candidate aptamers can simply be ignored, or the population can be subtracted or removed from the pool. Removal can be achieved, for example, by hybridization on a solid support containing probes that are complementary to a large number of sequences. Removal is also done by appropriately programming the DNA synthesizer to generate a pool of new sequences with all of the desirable features that imply "fitness", with the exception of sequences that bind to a high amount of protein. It may be an example of a reasonable intervention in that it excludes certain candidate sequences from subsequent iterations. This new pool is then advanced to a subsequent round of selection under the general scheme involving the present invention.

  Next, only the panel of candidate sequences with the least amount from the first step is noted and can be used subsequently. Thus, the quantitative power of the sequencer can focus on very small amounts of candidate sequences and, alternatively, proteins that are present in low abundance in the protein mixture. This is currently not possible with either conventional MS (mass spectrometry) based techniques or SELEX.

  Iterative subtraction of candidate sequences that are highly abundant should allow querying of lower abundance candidate sequences. To ensure that any binding is selective and does not just happen by chance, a known amount of known aptamer sequence is added to the candidate sequence and the known aptamer is present and binds in a known amount The sequence is operated on a mixture of proteins, including proteins. Alternatively or additionally, low dispersion across multiple replicates should mean that the query candidate sequence is highly likely to bind (stick) to the protein.

  Thus, the method of the present invention considers one of the major problems of proteomics (ie, how to deal with and address highly abundant proteins). This is accomplished by rationally excluding candidate sequences that bind to these proteins. These proteins are ignored in subsequent rounds of measurement and selection. The mutation / crossover process can cause the reappearance of several candidate sequences that bind to high abundance proteins, but these sequences are also rational in each iteration of the process as previously described. Can be eliminated. Careful selection of small amounts of candidate sequences (with strong sequence characteristics and low variance on their measurement between repeats) allows the population (or subset) of aptamers to bind specifically and sensitively to small amounts of protein It should allow convergence to the top.

  In further embodiments, the abundance range of candidate sequences in each iteration can also be monitored by counting candidate sequences and comparing them to a control population. This can be further used to improve fitness-based selection. As an example, during the first round of selection, the population of selected sequences is compared to that present in the initial unscreened library. Significant changes in the sequence composition and statistical prevalence of specific sequences (perhaps consistent with some models) can indicate relative success. If the first few rounds of selection do not produce a candidate sequence pool that is significantly different in composition from the initial population, the pool and experiment can be judged to have failed.

  It should be noted that once a candidate aptamer sequence or library of candidate aptamer sequences has been identified by the method of the present invention, the sequence must be verified as an aptamer. This can be achieved by replicating the binding and quantification of candidate sequences against multiple copies of the same original sample or across multiple different samples. If replication between samples is insufficient or the variability is high, the candidate sequence is probably not specific for one target. A suitable method for such validation is described in co-pending European Patent Application No. 070200498.8.

  The invention will now be further described by way of non-limiting examples.

Example 1
In this example, for comparison with SELEX, a set of aptamers is searched for a single protein present in solution or immobilized.

  A population of polynucleotide sequences such as DNA or RNA is expressed as a motif or pattern. For example, the sequences GGCT and CCGA can be represented by one pattern GGC (A / T), where “/” means “or”. An advanced set of IUPAC single letter codes exists to represent sequence patterns that allow multiple displays of sequences with emphasized commonality (20).

  In this example, DNA aptamer motifs are selected and a semi-random library is synthesized using known techniques (11). This is called a “candidate library”. The diversity of the library can be within any possible range of synthesis. In this example, reasonable prior knowledge is used to assist DNA sequences with specific secondary structure by self-similarity and annealing. A control library is also constructed from purely random sequences of the same length. This is called a “random library”.

  Next, as embodied in the conventional SELEX scheme, both libraries are screened against a single protein under selected conditions such that unbound aptamer sequences are discarded.

  Surviving aptamer DNA sequences (ie, protein bound sequences) are arrayed with a next generation DNA sequencer of appropriate resolution and dynamic range after unbinding the aptamer from the protein. This procedure is replicated to improve the statistical significance of the resulting measurement.

  All molecules on the array are sequenced and counted so that aptamers present in the surviving library are identified and counted. The number of sequences present on the array should represent the proportion of molecules in the originating population according to normal statistical sampling procedures. The accuracy of the measurement and any statistical sampling issues (especially with the rarer aptamers) can be confirmed by measuring the variance from each library replica.

  A differential comparison of the first iteration of the selected aptamer is performed against the initial unselected library. This also allows for successful library synthesis and evaluation of efficiency / diversity. Similarly, a differential analysis is performed on its unselected counterpart of a purely random library of aptamers. If the selected / exposed semi-random library is not significantly different from the non-exposed / random library, it can be concluded that aptamer selection is not appropriate and the library is not promising. At this point, the trial of this library can be stopped and another library with different characteristics at the sequence level can be synthesized.

  If the candidate library is sufficiently different from the unexposed candidate library and / or random library after the first iteration, the library will be used and further developed.

From an initial aptamer pool of any size, samples ranging from 10 ⁷ to 10 ⁹ aptamers will be sequenced on the array. Future generations of sequencers will enable higher dynamic ranges. Given the initial candidate library diversity and the results of sequencing and counting (and controls) in previous steps, the quality of the aptamer motif can be assessed after the first iteration. As described, this aptamer population can be advanced if deemed promising, or alternatively, a new initial candidate motif library can be generated. For example, if the distribution and variance of surviving aptamers are very different from the initial library as measured from controls, such libraries are preferred for progression.

  The first iteration provides data from experimental conditions and controls.

  In Table 1, calculate the bioinformatics properties derived from each aptamer sequence, its amount and variance, the secondary structure and other sequences, and the overall measure of sequence fitness via the aforementioned “fitness function” Can do. In use, the table has thousands or tens of thousands of entries, one row for each unique aptamer sequence. The column contains the aggregated measurements for that aptamer sequence. The table shown is for illustrative purposes only.

  Fitness calculations from multiple scores are well described in the literature on genetic algorithms (14, 15), and can include, by way of example only, multiple probability Bayes combinations. Any appropriately computable fitness measure can be used based on either the measured abundance of the sequence or the sequence itself (and any inferable properties). In the simplified example provided in Table 1, the motif GCT is clearly preferred, and in particular, it is shown that GCTG is heavily weighted by its amount relative to the control. In this way, each candidate aptamer sequence can be aggregated by application of fitness measurements and assessed for its potential as an aptamer. The score obtained as a measure of fitness is used to assess whether a given candidate sequence will survive the next iteration, allowing the sequence to grow (cross) or change its sequence composition It can also be used to determine whether to perform a mutation process. After applying these processes, other new aptamer sequences can be derived in addition to the original.

  The low variance between replications can also be an example of a statistic that demonstrates the specificity of a given motif or sequence for a protein. Thus, a duplicate of experimental measurements can be used to improve the derived statistical measurements illustrated in Table 1. An important aspect of the present invention is that appropriate statistics can be derived by experimentation and learning.

  Based on the information gathered at this stage, a new synthetic aptamer population can be generated based on measured statistics, bioinformatics and rational analysis from first generation results. In this example, purely for illustration, the library is re-synthesized on a DNA synthesizer rather than conventional amplification as used in SELEX. This resynthesis can include programming cross-combination of motifs derived from the aggregation of adaptive sequences, as shown in Table 1, and some new sequence random mutations at specific positions rather than other positions. . In this way, sequence diversity within the aptamer population is maintained at appropriately diverse levels in the initial iteration. In accordance with conventional genetic processes, subsequent new populations (Table 1) from the most adapted individuals of the previous population should also “adapt” or “more adapt”. As population generation fitness converges, intervention in population diversity will decrease. By using rational design, mutation and crossover procedures, the diversity of candidate aptamers can be maintained in the initial population. This ensures that the final solution does not encounter premature “minimum” or other search termination conditions (15) as described in other proven embodiments of such algorithms. become.

  The new candidate library (nth) is now exposed to the protein under study. Surviving bound aptamers are eluted, re-arrayed on the sequencer, and the general evaluation procedure described above is repeated.

  The previous sequencing iteration (n-1) of the candidate library can be a forwarded control as the new population is sequenced and counted, for the current iteration (n) that replaces the values in Table 1. The resulting table results. It can be seen that certain sequences rapidly dominate and are reflected in abundance and variance relative to the previous iteration (n-1), as well as during replication of the current iteration. Similarly, this tendency may not be seen. In the latter scenario, the number of “mutations” (random changes to the aptamer during synthesis) can be increased to promote the evolution and diversity of the population of sequences under study. In any case, the asymptotic convergence of the population towards a stable set of solutions (ie aptamers) can be tracked and optimized if necessary and as needed. Such a principle also ensures that the population can maximize the rate of climb towards a plateau with a good fitness and composition.

  An iteration of the candidate aptamer population is performed until a stable population of candidate solutions (aptamers) is achieved. In later iterations, once the superiority of a particular sequence motif becomes clear, it may be necessary or desirable to amplify the sequence between repeats rather than re-synthesize the sequence. Alternatively, synthesis and amplification rounds can be used alternately.

  The resulting sequence is a strong aptamer for the protein under study and is expected to find multiple similarly viable aptamers. Each viable aptamer may or may not be structurally different at the sequence level.

  Once one or more suitable aptamers have been identified, the aptamers can then be chemically modified, for example using nucleotide analogs, to increase stability and other desirable properties. The population can then be rescreened using the same or other procedures to confirm the specificity and other characteristics of each aptamer.

  Thus, for a given protein, a library of sequences shown to act as aptamers for that protein can be constructed. Most importantly, a population of strong solutions that did not necessarily exist in the initial candidate aptamer library because changes to the composition of the sequences were made possible during both random and rationally derived iterations. Can be found.

(Example 2)
While the above examples relate to the generation of aptamer libraries for a single known protein, the process can also be used for a mixture of known proteins in vitro. This is because the ratio of proteins to possible aptamers is very high. In this case, motif diversity is important at the sequence (implicitly structure) level. The initial population of diverse but functional sequences is therefore an important feature of the present invention.

  In this example, multiple known proteins are isolated for aptamer selection. The method proceeds as in Example 1 except for the purpose of selecting and managing multiple candidate aptamer libraries in parallel. This is accomplished by grouping aptamer libraries in one candidate library, or by including aptamers for all proteins under study starting with one sufficiently diverse candidate library. In accordance with the method described in Example 1, additionally, after the initial iteration, aptamers are grouped or clustered in the manner (16) already described in the genetic algorithm, and a plurality of concepts as shown in Table 1 A typical table. In Example 1, the grouping can be based on “clustering” or other methods that use the measured features embodied in Table 1. In this case, the inter-selection intervention functions to maintain a conceptually distinct table (representing a subpopulation of aptamers) that abstractly represents the mixture of proteins under study. In its simplest embodiment, this is a form of multiplexing.

  At the end of the selection process, aptamers specific to the protein in the mixture under study will be obtained. In this case, it may not be clear which aptamer subset binds to which protein. This can be resolved later using other experimental techniques, or may not be necessary. In the latter case, it is possible to generate and use “anonymous” aptamers for known proteins if the aptamer and protein are to be placed and used together.

  Alternatively, prior knowledge of the specificity of a particular aptamer sequence motif for a particular protein structure can be used to suggest which aptamer matches which protein.

  In both Examples 1 and 2, the specificity of aptamers identified for a given protein to those from which they were selected should be demonstrated as an additional step. This additional step can also be accomplished using next generation sequencing by exposing the aptamer to a complex biological sample containing a known amount of the selected protein. Aptamer measurements should not be confused by non-specific binding to proteins present in samples where they are not selected. This can be examined using counting statistics generated by arraying and sequencing of exposed aptamers.

(Example 3)
Ideally, the specificity of the aptamer for each protein is incorporated into the selection / generation process. This is not possible with SELEX schemes and schemes for single proteins or small subsets.

  In this example, the method proceeds as described in Example 1. However, in this example, the aptamer library is exposed to a native complex biological sample. This experiment is replicated in large numbers.

  First, aptamers that are very much represented on the array will be seen. Some or all of the well-represented aptamers can also have a high degree of dispersion between replicates. In the former case, it can be logically concluded that the aptamer binds a high amount of protein. If their variance is high, it is reasonable to conclude that aptamers lack specificity.

  In the next round of synthesis, these abundant or dispersed aptamers are excluded. The sequencing array is then outlined in Example 1 until it contains a diverse set of sequences that have a low relative representation between replicates (perhaps over some controls) and a very low variance. As described, the method of the present invention is performed. In this case, aptamers that bind a high amount of protein in the sample and / or aptamers with low specificity are effectively excluded.

  The subsequent aptamers can be used individually or in groups to repeat the method as set out in Example 2 using a mixture of known proteins. The identified aptamer sequence can provide a “motif” that is used to generate the semi-random library required in simpler examples. In this way, the library will be pre-filtered for sensitivity and specificity (a common cause of failure of antibody-based methods).

  In a further step, the population can be collectively continued to identify multiple aptamers for unknown low abundance proteins using the general scheme of Example 2. In this case, the protein to which the aptamer is bound is unknown. Protein identification will be elucidated in a further step using other techniques. However, it is conceptually possible to develop a set of “unnamed” aptamer probes for unknown low abundance proteins.

  This does not limit their usefulness as they can then be placed to measure these proteins in many samples. Aptamers here are surrogates of these proteins, and aptamers (ie proteins) that vary statistically significantly can therefore be measured between samples (ie find biomarkers). Achieving this, with the appropriate aptamer in the palm, can then be extracted and used to identify the surrogate protein. In this way, at first, like finding aptamers to complex mixtures of unnamed low-abundance proteins, placing them in biomarker discovery, finding significant aptamers, and current state of the art Rather, it is possible to identify the protein as a final step.

(References)
1. “A programmable biomolecular computing machine with bacterial phenotype output.” Kosoy E, Lavid N, Soreni-Harari M, BioC ", July 23, 2007, 8 (11), pages 1255-60. “DNA molecule supplies a computing machine with both data and fuel.” Benenson Y, Adar R, Paz-Elizur T, Livneh Z, Shapiro “Shapiro” 2. Proc Natl Acad Sci USA ”March 4, 2003, 100 (5), 2191-6. “Solving satisfiability problems using a novel array-based DNA computer.” Lin CH, Cheng HP, Yang CB, Yang CN “Biosystems” July-August 2007 90 (1), pages 242-252. “DNA computing using single-molecule hybridization detection.” Schmidt KA, Henkel CV, Rozenberg G, Spaink HP, “Nucleic Acids Research 9” May 23, 32 (17), 4962-8. “Fast parallel molecular algorithms for DNA-based computation: factoring integrators.” Chang WL, Guo M, Ho MS “IEEE Trans Nanobioscience”, June 2005, April (2) pp. 149-63 “Functional RNA microarrays for high-throughput screening of antiprotein receptors.” Collet JR, Cho EJ, Lee JF, Levy M, Hood A, Hood J Ellington AD “Anal Biochem”, March 1, 2005, 338 (1), pp. 113-23. “Nucleic acid evolution and minimization by nonhomogenous random recombination.” Bittker JA, Lo B (Le BV), Liu DR “Nat Biotechnol”, October 24, 2010 (10) Page 9 8. “Aptamers come of age-at last”. 8. Bunka DH, Stockley PG "Nat Rev Microbiol" August 2006, 4 (8), 588-96. “Aptamers: molecular tools for analytical applications.” Myral T, Cengiz Ozalp V, Lozano Sanchez P, Mir M, Katakis I, Osalivan (O 'ulli) “Analytical and Bio-Analytical Chemistry” June 21, 2007 10. “Aptamers as tools for target validation. Blank M, Blind M,“ Curr Opin Chem Biol ”, August 2005, 9 (4), 336-42. Page 11. “Methods developed for SELEX. Gopinath SC” Analytical and Bioanalytical Chemistry, January 2007, 387 (1), 171-82. “Aptamers-based assays for diagnostics, environmental and food analysis.” Tombelli S, Minunni M, Masini M, “Biomolecular Engineering, June 2006” Pages 191-200. “SELEX-A (r) evolutionary method to generate high-affinity nucleic acid ligands.” Stoltenburg (R), Reinemann (C), Strehlitz (B) “Bio-Molecular Engineering” (200) Moon, 24 (4), pages 381-40314. “Genetic Algorithms in Search, Optimization, and Machine Learning.” Goldberg, DE Addison-Wesley USA, 1989, 15. “Genetic Algorithms + Data Structures = Evolution Programs.” Michalewicz, Z, 3rd edition Springer, 1996. “Evolution Algorithms in Combinatorial Optimization,” H. Muhlenbein, M. Gorges-Schleuter, O. Kramer, “Parallel Computing 7” (1988), pages 65-88. 17. “Complex SELEX again target target: stochastic computer model, simulation, and analysis”. Chen CK “Comput Methods Programs Biomed”, September 2007, 87 (3), 189-200. “Elucidation of the Small RNA Component of the Transscript.” Lu et al. “Science 2” September 2005, pages 1567-1569 19. Qian WJ et al. "Mol Cell Protomics" (2006), 5 (10), pp. 1727 to 1744. http: // www. chem. qmul. ac. uk / ibummb / misc / naseq. html
21. “The potential and challenges of nanopore sequencing,” Branton et al., “Nature Biotechnology” (2008) 26, 1461-1153.

Claims (30)

A method for the identification of one or more aptamers against at least one target molecule comprising:
a) contacting a pool of candidate polymer sequences with at least one target molecule;
b) splitting the unbound sequence from the sequence specifically bound to the target molecule;
c) dissociating the sequence target complex to obtain a mixture enriched in ligands of the sequence;
d) assigning each sequence obtained in step c) a measure of the aptamer potential of the sequence (fitness function);
e) using the measurements of step d) to determine the aptamer potential of the ligand enriched mixture;
f) using the information obtained in step e) to allow the evolution of some or all of the sequences obtained in step c) to generate a new sequence mixture;
g) repeating steps a) -f) with the newly generated candidate aptamer pool until the total aptamer potential of the candidate pool reaches a plateau,
The method wherein the sequence present in the final pool is an optimal aptamer to at least one target molecule.   The method of claim 1, wherein the at least one target molecule is a protein.   The method of claim 2, wherein the at least one target molecule is a single isolated protein.   4. A method according to claim 1, 2 or 3, wherein it is known what one or each target molecule is.   5. A method according to claim 2 or 4, wherein the plurality of proteins are interrogated in a mixture of proteins.   6. The method of claim 5, wherein the at least one protein is a known protein present in a mixture of proteins.   The method according to claim 5 or 6, wherein the protein mixture is derived from a biological sample.   8. The method of claim 7, wherein the biological sample is a body fluid.   9. The method of claim 8, wherein the body fluid is blood or derived from blood.   The method according to claim 9, wherein the body fluid is serum or plasma.   The method according to claim 1, wherein the polymer sequence is a polynucleotide.   12. The method of claim 11, wherein the polynucleotide sequence is DNA, RNA, PNA (peptide nucleic acid), or variants or combinations thereof.   13. A method according to any one of claims 1 to 12, wherein the polymer is between 30 mer and 60 mer.   14. The method of claim 13, wherein the polymer is 40mer.   15. A method according to any one of the preceding claims, wherein the potential for aptamers is measured by quantification of candidate sequences in a mixture enriched for ligands.   16. The method of claim 15, wherein quantification is performed by sequencing at least a portion of each candidate sequence.   The method of claim 16, wherein sequencing is performed on a single molecule array or a clonal single molecule array.   18. A method according to any one of claims 1 to 17, further comprising arraying sequences in the mixture enriched with ligand onto the surface prior to step d).   19. The method of claim 18, further comprising amplifying the arrayed sequence.   20. The method of any one of claims 1-19, wherein the aptamer likelihood measure further comprises one or more measured properties, calculated properties, or bioinformatics properties.   21. The method of claim 20, wherein the bioinformatics characteristics include secondary structure prediction, tertiary structure prediction, self-similarity, information complexity, similarity to known aptamer sequences, sequence motifs, or combinations thereof.   6. The ligand enriched sequence, having a statistically significant aptamer potential when compared to the candidate sequence population under study, is advanced from step d) and e) to step f). The method according to any one of 1 to 21.   22. A ligand enriched sequence having the potential of aptamers classified in the average or upper percentile range is advanced from step d) and e) to step f). The method according to item.   24. A method according to any one of claims 1 to 23, wherein the ligand enriched sequences with the potential for statistically insignificant aptamers are removed from the candidate pool.   25. A method according to any one of claims 1 to 24, wherein unbound candidate sequences are eluted from the candidate pool and discarded.   26. The method of any one of claims 1-25, further comprising removing a numerically dominant sequence from the candidate pool.   27. The method of any one of claims 1-26, further comprising obtaining the complete sequence of candidate aptamers present in the final pool.   28. A method according to any one of claims 1 to 27, wherein a new candidate aptamer pool is designed using sequences with high aptamer potential or motifs derived from such sequences.   29. A sequence and / or motif having a high probability of an aptamer is used to affect random changes and / or recombination of that sequence that indicate a high probability of the aptamer. The method according to one item.   30. The method of any one of claims 1-29, further comprising modifying the candidate sequence to increase stability and / or binding ability.